How Makers Build
Check my stackPlan my stack
Menu

Helicone

LLM observability platform and gateway that logs, monitors and helps debug requests across OpenAI, Anthropic and other providers (acquired by Mintlify in March 2026).

In maintenance · Mar 2026: Acquired by Mintlify in March 2026; Helicone says the service stays live in maintenance mode — security fixes, new model support and bug fixes, but no new product direction. Source ↗

Visit helicone.ai ↗Pricing & calculatorAlternatives
Works with AI agents:llms.txt
Ask your AI about this, with this page as the source:ChatGPT ↗Claude ↗Perplexity ↗

Helicone logs every LLM request your app makes, with its cost, latency, tokens and errors, so you can see spending per user or feature, find failing calls and reproduce them. It also acts as a gateway, which lets it add behavior to those requests on the way through.

The main model is a proxy: you point your existing OpenAI-style client at a Helicone URL and add a Helicone key header, and it forwards the call to the provider and records it. Its AI Gateway goes further, exposing 100+ providers behind one OpenAI-compatible endpoint, either with your own provider keys or with Helicone credits. If you don't want it on the request path, an async logger based on OpenLLMetry sends logs after the fact instead, with fewer features.

Through the proxy you get sessions for tracing multi-step agent runs, custom properties and user metrics, response caching, custom rate limits, retries and provider fallbacks, prompt management, and datasets for evaluation or fine-tuning. The cloud proxy runs on Cloudflare Workers; you can also self-host it with an all-in-one Docker image, Kubernetes or a manual install.

In March 2026 Helicone was acquired by Mintlify and said its service would stay live in maintenance mode, meaning security fixes, new models and bug fixes rather than new product direction. The async integration gives up caching, retries and rate limiting.

Where it fits

LLM Observability & Evals · Proxy loggingGetting request logs, costs and latency by changing one base URL, with no SDK instrumentation.Trade-off: Proxy logging sees individual requests well but agent steps and eval workflows less deeply; acquired by Mintlify in March 2026, so check its roadmap before building on it.

How Helicone itself is built

28 tools, from its own code, website and Product Hunt page.

Amazon BedrockAmazon S3Better AuthClaudeClaude CodeClickHouseCloudflareExpressFastAPIFastifyLangChainLoopsMinIONext.jsNodemailerOpenAI APIPostgreSQLPostHogReactResendSentrysharpStripeSubstackSupabaseTogether AIVercelVercel Web Analytics

Who uses it

13 makers' products, each linked to the source that shows it.

The maker says so 9Customer story 5How evidence is collected →
TuskTurn your production traffic into tests

“Helicone makes it easy for our team to monitor our LLM requests and debug performance issues. We're a big fan of their UX!”

LLM Observability & EvalsMaker says so +1 · source ↗
Pretty PromptChrome extension that rewrites text into more structured prompts for AI chatbots such as…

“Helps us debug AI, and have clarity on AI analytics, costs, and latency. We've also been able to save quite a lot of credits because of its caching functionality!”

LLM Observability & EvalsMaker says so · source ↗
SpiralScale yourself by automating 80% of your repetitive tasks

“Observability, cost tracking and diagnosing how an LLM works is pretty hard without a tool like Helicone. It was a trusty tool in the making of this V2”

LLM Observability & EvalsMaker says so · source ↗
AlaiAI tool that generates and designs presentations, social posts, infographics and ads fro…

“Cannot live without this one.”

LLM Observability & EvalsMaker says so · source ↗
CodebuffBetter code generation than Cursor, from your CLI

“Helicone just works right out of the box – really helpful for us to dig into user issues and understanding how much money we're burning on LLM API calls.”

LLM Observability & EvalsMaker says so · source ↗
PersanaPersana helps sales teams find prospects, enrich them with data from many sources, track…

“Helicone AI offers open-source observability tools tailored for developers working with LLMs. It simplifies debugging and optimization, providing valuable insights into AI model performance.”

LLM Observability & EvalsMaker says so · source ↗
Image AllySet it and forget about it AI image alt tags and captioning

“I found Helicone in the middle of development, and it has been awesome. It gives me extremely useful and detailed insights on usage, costs, and response times, with just a couple of lines of code.”

LLM Observability & EvalsMaker says so · source ↗

What makers say

6 makers on why they use Helicone, in their own words on Product Hunt.

I found Helicone in the middle of development, and it has been awesome. It gives me extremely useful and detailed insights on usage, costs, and response times, with just a couple of lines of code.
Image AllySep 2026 ↗
Helicone AI offers open-source observability tools tailored for developers working with LLMs. It simplifies debugging and optimization, providing valuable insights into AI model performance.
PersanaSep 2026 ↗
Helps us debug AI, and have clarity on AI analytics, costs, and latency. We've also been able to save quite a lot of credits because of its caching functionality!
Pretty PromptSep 2026 ↗
Helicone just works right out of the box – really helpful for us to dig into user issues and understanding how much money we're burning on LLM API calls.
CodebuffSep 2026 ↗
Observability, cost tracking and diagnosing how an LLM works is pretty hard without a tool like Helicone. It was a trusty tool in the making of this V2
SpiralSep 2026 ↗

Loved and watch-outs

Themes that recur in makers' words and Hacker News comments, each linked to what it summarises, with how Product Hunt tags its reviews.

Most loved
  • Setup takes a couple of lines of code and works as a proxy, so it suits stacks outside Python too. PHHN
  • Custom properties attribute LLM cost, latency and usage to individual customers or features. helicone.aiPH
  • Request logs and session views make it practical to debug issues coming from real users. PHhelicone.aiHN
  • Built-in response caching cuts provider spend. helicone.aiPH
Watch-outs
  • Acquired by Mintlify in March 2026; some developers report the product has since moved to maintenance mode. helicone.aiHN
  • Per-run cost breakdowns need manual labelling and custom SQL, since views are retrospective. HNHN 2

Alternatives to Helicone

All alternatives by situation →
LangfuseTracing, prompt management and evals in one tool you can run on your own server for free.Open sourceFree tier33 makers · LLM observability & evalsCompare with Helicone →
LangSmithApps built on LangChain or LangGraph, where tracing works with almost no setup.Free tier11 makers · LLM observability & evalsCompare with Helicone →
BraintrustEval-driven work — scoring outputs and comparing prompts and models side by side in experiments.Free tier5 makers · LLM observability & evalsCompare with Helicone →
Arize PhoenixOpenTelemetry-based tracing and evals you can start locally in a notebook, then self-host or move to its cloud.Self-hostableFree tier
OpikTracing, evaluations and prompt optimization in an Apache-licensed platform you self-host with Docker or Kubernetes, or use on Comet's cloud.Open sourceFree tier
PromptfooRunning prompts and models against test cases from the command line or CI, including red-team tests for prompt injection and data leaks.Open sourceFree tier
DeepEvalWriting LLM evals as pytest-style tests in Python, with ready-made metrics for hallucination, faithfulness and answer relevancy.Open sourceFree tier
RagasScoring RAG pipelines — how faithful answers are to the retrieved context and how relevant that context is — and generating test questions from your documents.Open sourceFree tier

On Product Hunt, people weigh it against: Langfuse, Eden AI, LiteLLM.

Questions makers ask about Helicone

What is the status of Helicone after the Mintlify acquisition?

Helicone says its services will stay live for the foreseeable future in maintenance mode, which it defines as continuing security updates, new model support, and bug and performance fixes. source ↗

Should I use the proxy or the async integration?

The proxy is the easy setup and unlocks gateway features like caching, retries and custom rate limits, but it sits on the request path. Async logging keeps Helicone off the critical path so an outage on their side can't affect your app, at the cost of those gateway features. source ↗

What happens to my requests if Helicone has a problem?

The proxy applies no extra logic unless you enable features via headers, and its error handling falls back to plainly forwarding the request. Logs are sent to a separate pipeline after the full response is returned to you. source ↗

Can I self-host it?

Yes. Helicone documents a single all-in-one Docker image for quick or small setups, plus Kubernetes and manual installation; self-hosting support is through their Discord unless you have an enterprise agreement. source ↗

Do I need accounts with every model provider?

Not with the AI Gateway. You can buy Helicone credits and call 100+ providers through one OpenAI-compatible endpoint, or bring your own provider keys and use it for observability only. source ↗

What are the free plan's limits besides price?

The Hobby plan keeps data for 7 days, allows one seat and one organization, and caps ingestion and API access rates; longer retention and higher limits come with paid plans. source ↗

Is it SOC 2 or HIPAA compliant?

Helicone lists SOC 2 Type II and HIPAA compliance on its Team plan and above, with SAML SSO on Enterprise. source ↗

Is Helicone free?

Yes — there is a free tier a small product can run on; paid use starts at $79/mo. source ↗

Is Helicone open source or self-hostable?

Open source, and you can self-host it. source ↗

Can AI coding agents work with Helicone?

It serves an llms.txt docs index.

Who uses Helicone?

13 makers' products we track, each with a source. source ↗